Clinical Biomechanics 28 (2013) 831–845

Contents lists available at ScienceDirect

Clinical Biomechanics

j ourna l homepage: www.e lsev ie r .com/ locate /c l inb iomech

Review

Biomechanical characteristics of peripheral diabetic neuropathy:A systematic review and meta-analysis of findings from the gait cycle, muscle activityand dynamic barefoot plantar pressure

Malindu Fernando a,b,⁎, Robert Crowther b, Peter Lazzarini c,d, Kunwarjit Sangla e, Margaret Cunningham a,Petra Buttner f, Jonathan Golledge a,g

a Vascular Biology Unit, Queensland Research Centre for Peripheral Vascular Disease, School of Medicine and Dentistry, James Cook University, Townsville, Australiab Movement Analysis Laboratory, Institute of Sports and Exercise Science, James Cook University, Townsville, Australiac Allied Health Research Collaborative, Metro North Hospital and Health Service, Queensland Health, Australiad School of Clinical Sciences, Queensland University of Technology, Brisbane, Australiae Department of Internal Medicine, The Townsville Hospital, Townsville, Queensland, Australiaf School of Public Health and Tropical Medicine, James Cook University, Townsville, Queensland, Australiag Department of Vascular and Endovascular Surgery, The Townsville Hospital, Townsville, Queensland, Australia

⁎ Corresponding author at: Vascular Biology Unit, QuPeripheral Vascular Disease, School of Medicine, Jame4814, Australia.

E-mail address: malindu.fernando@my.jcu.edu.au (M.

0268-0033/$ – see front matter © 2013 Elsevier Ltd. All rihttp://dx.doi.org/10.1016/j.clinbiomech.2013.08.004

a b s t r a c t

a r t i c l e i n f o

Article history:

Received 30 May 2013Accepted 20 August 2013

Keywords:Diabetic peripheral neuropathyBiomechanicsGaitDiabetes complicationsType 2 diabetesType 1 diabetesPlantar pressureElectromyographyMovement analysisDiabetic footDiabetes mellitusMeta-analysisSystematic review

Background: Diabetic peripheral neuropathy is an important cause of foot ulceration and limb loss. This system-atic review andmeta-analysis investigated the effect of diabetic peripheral neuropathy on gait, dynamic electro-myography and dynamic plantar pressures.Methods: Electronic databaseswere searched systematically for articles reporting the effect of diabetic peripheralneuropathy on gait, dynamic electromyography and plantar pressures. Searches were restricted to articles pub-lished between January 2000 and April 2012. Outcome measures assessed included spatiotemporal parameters,lower limb kinematics, kinetics, muscle activation and plantar pressure. Meta-analyses were carried out on alloutcome measures reported by ≥3 studies.Findings: Sixteen studies were included consisting of 382 neuropathy participants, 216 diabetes controls withoutneuropathy and 207 healthy controls. Meta-analysis was performed on 11 gait variables. A high level of hetero-geneity was noted between studies. Meta-analysis results suggested a longer stance time andmoderately higherplantar pressures in diabetic peripheral neuropathy patients at the rearfoot, midfoot and forefoot comparedto controls. Systematic review of studies suggested potential differences in the biomechanical characteristics(kinematics, kinetics, EMG) of diabetic neuropathy patients. However these findings were inconsistent and lim-ited by small sample sizes.Interpretation:Current evidence suggests that patientswith diabetic peripheral neuropathy have elevatedplantar

pressures and occupy a longer duration of time in the stance-phase during gait. Firm conclusions are hamperedby the heterogeneity and small sample sizes of available studies.

© 2013 Elsevier Ltd. All rights reserved.

1. Introduction

One of the many consequences of diabetes is the onset of diabeticperipheral neuropathy (DPN) (Shenoy, 2012). The prevalence of DPNranges from 13 to 68% in diabetes populations (van Dieren et al.,2010). Peripheral neuropathy affects the sensory,motor, and autonomiccomponents of the nervous system, manifesting as a loss of protectivesensation, intrinsic foot muscle dysfunction and anhydrosis of the foot

eensland Research Centre fors Cook University, Townsville

Fernando).

ghts reserved.

(Shenoy, 2012). These manifestations often lead to bony deformitiesand high plantar pressure areas which result in skin breakdown and ul-ceration (Boulton et al., 2005). It is believed that themajority of diabeticfoot ulcers develop as a result of the repetitive action of mechanicalstress (pressure) during gait, in the presence of peripheral neuropathyor loss of protective sensation (Armstrong et al., 2004). Lower-limb am-putations in people with diabetes are typically preceded by foot ulcera-tion, suggesting that better understanding of the mechanisms of ulcerdevelopment are of vital importance (Singh et al., 2005). This includesbetter understanding of the biomechanical components (Formosaet al., 2013).

It has been postulated that DPN-related changes in the lower limbsmay lead to functional gait variations; predominantly related to reducedrange of movement of joints, reduced active muscle power and changes

832 M. Fernando et al. / Clinical Biomechanics 28 (2013) 831–845

in gait mechanics (Andersen, 2012). The biomechanical changesresulting from DPN may translate to increased plantar pressures in thefoot, which contributes to the pathogenesis and development of foot ul-cers, especially in the forefoot (Van Deursen, 2004). In particular, thefirst metatarsophalangeal joint has been implicated as a site of biome-chanical dysfunction leading to elevated plantar pressures during gait,promoting ulceration at this site (Turner et al., 2007). Therefore, wehypothesised that reductions in spatiotemporal parameters, increasesin kinetics (specifically the vertical ground reaction force and joint mo-ments), and reductions in kinematics of the lower limb (evident as re-strictions in the sagittal plane) and altered dynamic electromyography(EMG) findings in those with DPNmay manifest from or contribute to-wards altered plantar pressure loading in this population (Cavanaghet al., 2000). Therefore, this systematic review and meta-analysisaimed to assess the effect of DPN on gait (spatiotemporal parameters,joint angular kinematic and kinetics), dynamic EMG (muscle activationand deactivation patterns) and dynamic barefoot plantar pressures(plantar foot pressures during gait). We sought case–control studiescomparing patients with DPN to those with diabetes mellitus withoutneuropathy (diabetes mellitus controls) (DMC) or healthy controls(HC).

2. Methods

2.1. Literature search strategy

Electronic databases (Ovid, CINAHL, PubMed, Scopus and GoogleScholar) were searched systematically by the first author for articlespublished between January 2000 to April 2012, reporting studies onDPN in the three biomechanical areas of gait, dynamic EMG and plantarpressure. The initial search was conducted in April 2012. An additionalsearchwas conducted in January 2013 to ensure that any further articleswere also assessed for inclusion prior to publication. No new articleswere found. Search resultswere restricted to articles published betweenJanuary 2000 and January 2013. Publications prior to the twenty firstcentury were not included to restrict the focus of the review to themost recent findings from studies which assessed gait using currenttechnology, which is more reliable and comprehensive. This is espe-cially true in relation to three dimensional joint angular kinematicanalysis which was introduced at around this time (Sutherland,2001, 2002, 2005). The following keywords and MeSH headingswere used:

1. Gait AND diabetes2. electromyograph* AND diabetes3. EMG AND diabetes4. biomechanic* AND diabetes5. kinematic AND diabetes6. plantar pressure AND diabetes7. (diabetes MeSH) AND 1# AND 2# AND 3# AND 4# AND # 58. (diabetic foot MeSH) AND 1# AND 2# AND 3# AND 4# AND # 59. (diabetic neuropathy MeSH) AND 1# AND 2# AND 3# AND 4#

AND # 5

2.2. Selection of studies

The titles and abstracts retrieved from the initial database searchwere screened by the first author utilising the question ‘Did the studyinvestigate one of the three biomechanical areas of interest?’ The fulltext was obtained for articles that remained relevant after the initialscreening. One of the authors then reviewed the full text for the final de-cision on inclusion utilising the entry criteria. All articles meeting theseinitial criteria had their full-texts retrieved and were then further eval-uated by two authors (MF and RC) using the inclusion and exclusioncriteria below. All studies meeting the exclusion criteria were removedfrom the review.

The inclusion criteria were:

1. Studies published between 2000 and 2012;2. Studies in the English language;3. Studies reporting findings in clearly identified DPN groups in com-

parison to a DMC and/or a HC group using eligible inclusion and/orscreening criteria;

4. Studies investigating barefoot walking. Barefoot investigations werechosen over shod as this was thought to provide insight into biome-chanical parameters without the influence of shoes;

5. Studies in adult populations (≥18 years old);6. Study reported findings for at least 1 outcomemeasure of interest in

the review.

Exclusion criteria were:

1. Any study investigating participants' gait, EMG or plantar pressurewhile wearing shoes, inserts or orthotic devices;

2. Any study which included current or past diabetes foot ulcer partic-ipants as a part of their DPN or DMC groups;

3. Studies that investigated movement on a treadmill;4. Studies where reported outcome measures were not comparable

with at least one outcome measure of interest and could not beconverted;

5. Studies where authors were unable to provide datasets or outcomevariables that were compatible for comparison (mean and standarddeviation, SD), in place of missing data.

2.3. Outcome measures

Studies were included in the review if they reported at least one ofthe following outcome measures:

1. Spatiotemporal—walking speed (m/s) with or without stride length(m);

2. Kinetics — reported findings on net moments of force (flexion andextension) for at least one lower limb joint (ankle, knee or hip)and/or reported ground reaction force at initial contact and/or toe-off as separate values;

3. Kinematics — reported range of motion (RoM) findings for at leastone lower limb joint (ankle, knee or hip) in both flexion and exten-sion directions;

4. EMG— activation and deactivation durations of any lower limbmus-cle during walking in % stance or % gait cycle;

5. Plantar pressure — reported on at least one site at the rearfoot ormidfoot or forefoot or in any other plantar location in either peakplantar pressure (MPP) or pressure time integral (PTI) or both.

2.4. Assessment of methodological quality of studies

Two assessors (MF and PL) independently evaluated the quality ofthe studies utilising a modified version of the quality assessment toolby Downs and Black (1998). The criteria within the tool which werenot applicable to the studies included in this review were omittedfrom the analysis (see Table 1). The total quality scores were reportedas an average score between the two assessors. As a simplified versionof the quality assessment instrument tool by Downs and Black (1998)was utilised, the original scoring system for the tool was scaledaccording to a total score of 18. Therefore, a score of≤7was consideredlow quality, 8–11 as fair quality and N11 as good quality.

2.5. Data extraction and reporting

Data extraction was performed by the first author with assistancefrom a statistician (PB) for data analysis. Data were entered into tablesfor ease of comparison and grouping of variables. Only studies that re-ported the outcome measures of interest were used in the statisticalanalysis that followed. Descriptive characteristics of participants (age,

Table 1Assessment of methodological quality of studies.

1

2

3

5a

6

7

10

11

12

13

16

18

20

21

22

25

27b

Total

Mean

Y

Y

Y

P

Y

Y

N

–

–

–

Y

Y

Y

–

–

Y

N

10

10

Y

Y

Y

P

Y

Y

N

Y

–

Y

–

Y

Y

Y

–

Y

N

12

13

Y

Y

Y

P

Y

Y

Y

Y

–

Y

–

Y

Y

Y

Y

Y

N

14

14

Y

Y

Y

P

Y

Y

Y

–

–

–

–

Y

Y

–

–

N

N

9

8

Y

Y

Y

P

Y

Y

N

Y

–

–

Y

Y

Y

Y

–

Y

N

12

12

Y

Y

Y

N

Y

Y

N

Y

–

Y

Y

–

Y

Y

–

N

N

10

10

Y

Y

Y

P

Y

Y

N

Y

–

Y

Y

Y

N

Y

–

N

N

12

12

Y

Y

Y

P

Y

Y

N

Y

–

Y

–

Y

Y

–

–

N

N

10

9

Y

Y

Y

P

Y

Y

Y

–

Y

Y

Y

Y

Y

Y

–

N

N

13

13

Y

Y

Y

P

Y

Y

Y

Y

–

Y

Y

N

Y

Y

–

N

N

12

12

Y

Y

Y

Y

Y

Y

N

Y

–

Y

N

Y

Y

Y

–

Y

N

13

13

N

Y

Y

P

Y

Y

N

Y

–

Y

Y

Y

Y

–

–

N

N

8

8

N

Y

Y

Y

Y

N

N

Y

–

Y

Y

Y

Y

Y

–

N

N

11

11

Y

Y

Y

Y

Y

Y

Y

Y

–

Y

–

Y

Y

Y

–

N

N

13

13

Y

Y

Y

Y

Y

Y

N

–

–

–

Y

Y

Y

Y

–

–

N

11

11

Y

Y

Y

Y

Y

Y

Y

–

–

Y

Ass

ess

or

2 (

PL)

Y

Y

Y

Y

–

–

N

13

12

Y

Y

Y

P

Y

Y

N

–

–

–

Y

Y

Y

–

–

Y

N

10

Y

Y

Y

P

Y

Y

Y

Y

–

–

Y

Y

Y

–

–

Y

N

12

Y

Y

Y

P

Y

Y

Y

Y

–

Y

–

Y

Y

Y

Y

Y

N

14

Y

Y

Y

P

Y

Y

Y

–

–

–

–

Y

Y

–

–

N

N

8

Y

Y

Y

P

Y

Y

N

Y

–

Y

Y

Y

Y

Y

–

Y

N

13

Y

Y

Y

N

Y

Y

N

Y

–

Y

Y

Y

Y

Y

–

N

N

11

Y

Y

Y

P

Y

Y

N

Y

–

Y

Y

Y

Y

Y

–

N

N

13

Y

Y

Y

P

Y

Y

N

Y

–

Y

Y

Y

–

–

–

N

N

8

Y

Y

Y

P

Y

Y

Y

–

Y

Y

Y

Y

Y

Y

–

N

N

13

Y

Y

Y

P

Y

Y

Y

Y

–

Y

Y

Y

Y

Y

–

N

N

13

Y

Y

Y

Y

Y

Y

N

Y

–

Y

N

Y

Y

Y

–

Y

N

13

Y

Y

Y

P

Y

Y

N

Y

–

–

–

Y

Y

Y

–

N

N

9

N

Y

Y

Y

Y

N

N

Y

–

Y

Y

Y

Y

Y

–

N

N

11

Y

Y

Y

Y

Y

Y

N

Y

–

Y

Y

Y

Y

Y

–

N

N

13

Y

Y

Y

Y

Y

Y

N

–

–

–

–

–

–

Y

Y

Y

Y

–

Y

–

N

N

12

Y

Y

Y

P

Y

Y

Y

Y

Y

Y

Y

N

11

Question

Ak

ash

i (2

00

8)

[1]

Ba

cari

n (

20

09

) [5

]

Ca

sell

i (2

00

2)

[9]

Go

me

s (2

011

) [1

8]

Gu

lde

mo

nd

(2

00

8)

[20

]

Me

lai

(20

11)

[23

]

Sa

cco

(2

00

9)

[29

]

Sa

ura

(2

01

0)

[30

]

Sav

elb

erg

(2

00

9)

[31

]

Sav

elb

erg

(2

01

0)

[32

]

Saw

ach

a (

20

09

a)

[33

]

Saw

ach

a (

20

09

b)

[34

]

Saw

ach

a (

20

10

a)

[35

]

Saw

ach

a (

20

12

b)

[36

]

Ucc

ioli

(2

00

1)

[46

]

Yav

uze

r (2

00

6)

[51

]

Ak

ash

i (2

00

8)

[1]

Ba

cari

n (

20

09

) [5

]

Ca

sell

in (

20

09

) [9

]

Co

mm

es

(20

11)

[18

]

Gu

lde

mo

nd

(2

00

8)

[20

]

Me

lai

(20

11)

[23

]

Sa

cco

(2

00

9)

[29

]

Sa

ura

(2

01

0)

[30

]

Sa

celb

erg

(2

00

9)

[31

]

Sa

celb

erg

(2

01

0)

[32

]

Saw

ach

a (

20

09

a)

[33

]

Saw

ach

a (

20

09

a)

[34

]

Saw

ach

a (

20

12

a)

[35

]

Saw

ach

a (

20

12

b)

[36

]

Ucc

ioli

(2

00

1)

[45

]

Yav

uze

r (2

00

6)

[41

]

Ass

ess

or

1 (

MF

)

Legend:Methodological quality of studies as assessed independently by (MF) and (PL) using a shorter end version of the quality assessment tool by Downs and Black (1998). 1 = (Y) Yes,0 = (N) No and (–) = unable to determine. The Total scorewas out of 18. Mean scores were the average of the two assessors' rating rounded down to the nearest integer. Ql; hypothesis,aims, objectives, Q2; relates to main outcomes described, Q3; patient characteristics, Q5; addressing confounders, Q6; description of main findings, Q7; random variabilityestimates, Q10; reporting probability values, Q11; representative of population (asked), Q12; representative of population (agreed), Q13; methods representative, Q16; mention ofdata dredging, Q18; use of appropriate statistical test, Q20; accurate outcome measures, Q21; internal validity (selection bias), Q22; recruitment time period, Q25; adjustment forconfounding, Q27; statistical power determined.a This question was scored on the basis of P (partial) = l, Y = 2 and N = 0 and the original scoring criteria was used.b This question was converted to a yes/no answer for ease of comparison.

833M. Fernando et al. / Clinical Biomechanics 28 (2013) 831–845

gender, body mass index, BMI), entry criteria for diagnosis of DPN, siteof participant recruitment, exclusion criteria used by study and diabetesduration of groups were recorded. Where data were missing orunreported, authors from the studies were contacted in an attempt toobtain or clarify results. Where authors did not reply, the studies wereexcluded from the review. The MOOSE guidelines for reporting meta-analysis of observational studies were utilised in the synthesis of this re-view (Stroup et al., 2000).

2.6. Statistical analysis

Where possible, data were transformed into standardised units ofmeasure for comparison and for statistical analysis. Means (weightedby sample size of the study) were calculated for the reported demo-graphic variables. Meta-analysis was carried out on individual outcomemeasures when more than three studies reported on the particularindividual outcome measure. Difference in mean values divided bypooled SD was used to compute effect size, utilising Cohen's d(Cohen, 1988). Heterogeneity of studies was calculated using theQ-statistic and I2 statistic. Results were reported as standardisedmean differences with 95% confidence intervals and P values. In ad-dition to this, the classic fail safe N was also computed, as this givesan estimation of studies needed to be published with a null effectto renounce the effects from the meta-analysis (Persaud, 1996). Forpurposes of analysis, a Cohen's d score of zero was interpreted asno difference in effect; a result of 0 to 0.2 was interpreted as asmall effect; 0.2–0.5 as a medium effect; and ≥ 0.8 as a large effect(McGough and Faraone, 2009). All statistical analyses were carriedout by a statistician (PB).

3. Results

3.1. Search yield

Fig. 1 outlines the process and results of each step of the literaturesearch. Overall, 1813 unique records were originally identified. Howev-er, 1800 articles were excluded for a variety of reasons, such as inappro-priate study design, use of inappropriate comparison groups, unsuitablemethods used in data capture, lack of neuropathy classification, missingdata, irrelevant data or because data were unable to be acquired fromauthors. Thus 13 articles remained eligible for inclusion. Three extra ar-ticleswere located by hand searching reference lists of included articles.Therefore, 16 articles were included in the review. Several studies re-ported on more than one focus area. Gait findings (spatiotemporalparameters, kinematics and kinetics) were reported in ten studies. Dy-namic EMG resultswere reported in three studies and barefoot dynamicplantar pressure in seven studies. Table 2 displays a summary of thecharacteristics of participants in included studies.

3.2. Study quality

Although there were minor differences in ratings between qualityassessors for studies, the overall agreement between quality assessorswas good. There were no studies which had a score ≤7 and thereforeseven studies were of fair quality (8–11) and nine studies were ofgood quality (N11) according to the quality assessment instrument(Downs and Black, 1998). The main difference between the studiesthat achieved a good quality compared to a fair quality was thereporting of actual probability values (i.e. P = 0.004) rather than ap-proximate values (i.e. P b 0.05) along with a more comprehensive list

Fig. 1. Flow-chart of all included and excluded studies.

834 M. Fernando et al. / Clinical Biomechanics 28 (2013) 831–845

of confounding variables. Additionally, the ‘good quality’ studies de-scribed the demographic and recruitment sites of the participants in de-tail and reported the populations of recruitment for groups as being thesame or different. This information is important for assessing the exter-nal and internal validity of studies. Themajority of ‘good quality’ studiesalso used ameans of adjustment for confoundingwith or without usingmultiple regression analysis.

None of the studies reported sample size calculations. All except twostudies had clear aims (Sawacha et al., 2012a, 2012b). All studies dif-fered in reporting of confounding variables, especially pertaining to bio-mechanical outcomes. Important confounders of relevance includeddiabetes duration, severity of DPN, presence of foot deformity, BMI, gen-der and presence of claudication pain or presence of peripheral arterialdisease (PAD) affecting gait. One study (Melai et al., 2011) did not re-port any major confounders. One study did not provide random esti-mates of variability in their manuscript, however this was found in thesupplementarymaterial (Sawacha et al., 2012a). Itwasdifficult to ascer-tain whether or not the recruited samples were representative of thesource population in most studies, however some studies stated the re-cruitment strategy clearly. Only one study commented on the numberof participants who accepted and rejected invitation for the study aspart of the external validity assessment (Savelberg et al., 2009). Lastly,only one study reported on time frames for recruitment, as it was a

part of a larger study (Caselli et al., 2002). For other studies this couldnot be determined.

3.3. Participant characteristics

Therewere 382 DPN participants (cases) in total from the 16 includ-ed studies. The mean group size was 25.5 and ranged from 8 to 76 par-ticipants. The age range of participants in the DPN group was 54 to69 years with a weighted mean age of 61 years. The majority (55%) ofsubjects were males with a BMI of 24 to 30 kg/m2 (weighted mean28 kg/m2). The weighted mean diabetes duration in the DPN groupwas 15.2 (range 12 to 27) years.

Studies utilised a variety of participant recruitment sources includ-ing community outpatient settings (8/16), hospital settings (3/16), vol-unteers (2/16), previous studies (1/16) or unspecified (2/16). Thirteenstudies utilised HC groups for comparison which were recruited on avoluntary basis from the community or through hospital staff. Ninestudies (9/16) usedDMCgroups usually recruited from the same settingasDPN patients. A summary of the recruitmentmethods can be found inTable 3.

Table 3 presents the screening criteria for the diagnosis of DPN andpopulation source of samples used in each study as well as additionalexclusion criteria used. A range of methods was used in the diagnosisof DPN in different studies. Eleven studies utilised a validated screeningtool to assess sensory neuropathy. The most commonly used was theMichigan Neuropathy Screening Instrument (MNSI). A few studiesused only clinical assessment (4/16) and one study utilised nerve con-duction testing to assess both motor and sensory neuropathy (Yavuzeret al., 2006). All studies disqualified patientswith previous or current di-abetes foot ulcers from inclusion in the DPN group and excluded thosewith additional orthopaedic and neurological conditions, rheumatolog-ical conditions and disabilities which produce walking constraints.Two studies excluded participants with PAD, assessed on clinical exam-ination or with ankle brachial pressure index (ABPI) values b0.85(Guldemond et al., 2008; Uccioli et al., 2001). Two studies did not spec-ify exclusion criteria (Caselli et al., 2002; Sawacha et al., 2009a).

The DMC group comprised of 216 participants with a meansample size of 24. The age of patients was 50 to 64 years with aweighted mean age of 57 years. The BMI of this group ranged from 25to 30 kg/m2 with a weighted mean of 28 kg/m2. The majority of partic-ipants were male (50%) and the total diabetes duration was lower thanfor the DPN group; ranging from 8 to 23 years and with a weightedmean of 14 years. The HC group comprised of 207 participants with amean sample size of 15.9. The age of participants ranged from 46 to68 years with a weightedmean age of 58 years. Themajority of HC par-ticipants were male (53%) with a BMI range between 24 and 29 kg/m2

andwith aweightedmeanBMI of 25 kg/m2. Thefindings from the stud-ies are reported in their respective sections below and in Tables 4–8.Meta-analysis was carried out for 11 separate gait variables (Table 9).Forest-plots of all significant meta-analyses can be found as additionalfigures (Figs. 2–9).

3.4. Spatiotemporal parameters

There was a marked difference in the walking speeds reportedamongst the three groups in different studies (Table 4). Three studies re-ported that DPN participants walked slower than HC subjects (Gomeset al., 2011; Savelberg et al., 2010; Sawacha et al., 2009b) and two stud-ies reported slower walking speeds in the DPN group compared to theDMC patients (Savelberg et al., 2010; Sawacha et al., 2012b). However,two studies reported that DPN participants walked faster than both HCand DMC groups (Sawacha et al., 2009a; Yavuzer et al., 2006).

Meta-analysis results combining data from studies for walkingspeed (DPN group vs. DMC group and DPN group vs. HC group) andstride length (DPN group vs. DMC group) between the three groupsdemonstrated no significant difference in walking speed and stride

Table 2Characteristics of study participants in included studies.

Cases Diabetes controls (DMC) Healthy controls (HC)

Study Focus area DPN(n=)

Age(years)

% male BMI(kg/m2)

Diabetes duration(years)

DMC(n=)

Age(years)

% male BMI(kg/m2)

Diabetes duration(years)

HC(n=)

Age(years)

% male BMI(kg/m2)

Akashi et al. (2008)a EMG 19 57.6 48 26.6 (4.2) 12.6 (5.3) No diabetes control group used 16 51.1 (8.3) 50 23.9 (2.9)Bacarin et al. (2009) PP 17 54.7 (7.8) 47 26.1 (4.6) 13.4 (8.2) No diabetes control group used 20 48.7 (9.4) 35 24.3 (2.6)Caselli et al. (2002) (9) PP 57 59.7 (11.7) 74 28.9 (5.8) 16.4 (10.9) 20 50.2 (16.2) 35 27.9 (4.1) 8.4 (10.4) No healthy control group usedGomes et al. (2011) (18) Gait and EMG 23 56 (8.0) 39 – 14.4 (6.5) No diabetes control group used 23 55 (8.0) 39 –

Guldemond et al. (2008) (12) PP 44 58.8 (10.4) 34 29.6 (6.5) N10 49 50.9 b 9.51 39 28.2 (14.8) N10 No healthy control group usedMelai et al. (2011) (23) PP 76 66.0 (7.2) – – – 33 62.8 (7.1) – – 19 68.1 (5.2) – –

Sacco et al. (2009) (29) PP 15 57 (6.0) 60 – N5 No diabetes control group used 16 46 (11.0) 31 –

Saura et al. (2010) (30) Gait 16 63 (3.89) 31 28.78 (2.88) 12.13 (1.26) 10 63 (3.92) 30 27.66 (2.66) 12 (1.15) 10 62 (3.77) 40 27.35 (2.5)Savelberg et al. (2009)# (31) PP & Gait 8 68.9 (6.3) 88 28.0 (3.2) 19.0 (13.6) 10 60.5 (6.9) 70 29.2 (3.7) 8.1 (0.6) 10 72.4 (6.0) 80 24.7 (2.9)Savelberg et al. (2010) (32) EMG and Gait 8 68.9 (6.3) 88 28.0 (3.2) 19.0 (13.6) 10 60.5 (6.9) 70 29.2 (3.7) 8.1 (0.6) 10 72.4 (6.0) 80 24.7 (2.9)Sawacha et al. (2009a) (33) Gait 26 63.2 (6.0) 58 25.6 (2.9) 22.1 (14.3) 21 63.8 (5.4) 70 26.3 (2.0) 17.2 (11.7) 20 59.0 (2.9) 65 24 (2.8)Sawacha et al. (2009b) (34) Gait 10 64.0 (6.8) 70 24.0 (2.8) – No diabetes control group used 10 61.8 (4.3) 80 24.1 (3.0)Sawacha et al. (2012a) (35) Gait and PP 12 62.0 (6.0) 75 25.2 (3.2) 26.7 (10.5) No diabetes control group used 12 60.3 (5.2) 83 24.1 (2.6)Sawacha et al. (2012b) (36) Gait 20 61.2 (7.7) – 26.8 (3.4) 13 (6.5) 20 56.53 (13.29) – 26.4 (2.5) 23.3 (13.7) 10 61.2 (5.0) – 24.4 (2.8)Uccioli et al. (2001) (45) Gait 19 53.7 (10.4) 52 27.0 (4.9) 19.4 (9.3) 27 52.7 (12.7) 70 25.3 (3.4) 15.1 (9.3) 21 56.6 (11.8) 61 25.0 (3.1)Yavuzer et al. (2006) (51) Gait 20 61.7 (8.5) 60 29.5 (4.5) 18.5 (5.4) 26 58.2 (9.5) 46 30.3 (5.9) 14.6 (8.7) 20 60.9 (5.9) 50 28.9 (5.1)Total (n = 16) 382

(mean 25.5)60.8 55.0% 27.7 15.2 216

(mean 24)56.5 50.1% 27.6 13.5 207

(mean 15.9)57.8 53.2% 25.1

Legend: Reported means and standard deviations (SD) of participants.DPN = diabetic peripheral neuropathy (cases) compared to DMC = diabetes mellitus control and healthy controls (HC). Study focus indicates whether the focus was gait (Gait), EMG or plantar pressure (PP) or a combination.Savelberg 2009/2010 has two publications from the same study, therefore only one was used (#) when calculating total sample size and demographic data.Total: Total sample size per group as well as averages weighted by sample size.

a This study did not report appropriate outcome measures for the inclusion of plantar pressure data.

835M.Fernando

etal./ClinicalBiomechanics

28(2013)

831–845

Table 3Characteristics of all included studies.

Study Recruitment source Neuropathy classification for inclusion Exclusion criteria

Sawacha et al. (2012a, 2012b) Cases fromoutpatient university clinic andcontrols were from hospital personnel,University of Padova.

A score of N3/15 on the MichiganNeuropathy Screening Instrument andcomprehensive neurologicalexamination.

History of ulcers or neurological disorders,orthopaedic problemsLower limb surgery, CV disease

Sawacha et al. (2009b) Anti-diabetic Unit, University Hospital ofPadova, Italy.

Anamnesis and clinical evaluation Not specified

Sawacha et al. (2009a) Anti-diabetic Unit, University Hospital ofPadova, Italy.

A score of N3/15 on the MichiganNeuropathy Screening Instrument.Comprehensive neurological examination.

History of ulcers or neurological disorders,orthopaedic problems, lower limbsurgery, CV disease

Savelberg et al. (2009, 2010)a Not specified Clinical examination with sensory testing,tendon reflexes and muscle strength inthe lower extremities.Well regulated metabolic controla

BGL 5–16 mmol/l during trialsa

Subjects with foot deformities or activeulcerationLimited mobility due to severe jointproblems, angina pectoris or cardiacfailure NYHA Class 2 or higherMyocardial infarction within 1 year,BMI N 35Orthopaedic or neuromuscular diseaseother than diabetic polyneuropathySeverely restricted mobility of lowerextremity joints (less than 10° of dor-sal flexion or less than 30° of plantarflexion)a

Saura et al. (2010) Cases were patients of the foot and ankle-group, outpatient centre, Faculty of Medi-cine, University of Sao Paulo. Controlswere community members.

Michigan protocol Neuroischemic diabetic foot, Charcot'sneuroarthropathyInability to mobilise.

Sacco et al. (2009) Cases were recruited through TheBrazilian Association for Assistance to Di-abetics. The control subjects were em-ployee volunteers from the academicdepartment.

Michigan Neuropathy score N6/15 Type 2diabetes with N5 year onset.10 g monofilament absence in 2 plantarareas of the footModified MNSI to assess toe deformities(additional)

Modified MNSI score of N4 (foot deformi-ty)Over 65 years ageHallux amputations or partial foot ampu-tations—except toesOrthopaedic problems of the lower limb.

Akashi et al. (2008) Volunteers At least 5 years post-onset of Type 2 dia-betes,At least 2 insensate areas of the foot to themonofilament and aminimal score of 6 onthe Michigan neuropathy score.Modified MNSI score of b4/6.

Age over 65 yearsPartial or total amputationCharcot arthropathy or any other footdeformity confirmed by radiology. Otherneurological disease, alcohol abuse,Presence of plantar ulcers, inability towalk.

Melai et al. (2011) Cases from MaastrichtUniversity Medical Centre, MaximaMedical Centre, Eindhoven andVeldhoven, St. Anna Hospital Geldrop andMaasland Hospital, Sittard.Controls from advertisement orparticipants from previous research

Standardised neurological examination. Severe CVD, renal dysfunction, intermit-tent claudication,Neurological disorders, RA, severe OA,Foot amputations or deformities,Foot ulcers.

Guldemond et al. (2008) Cases were selected from the outpatientclinic of the University HospitalMaastricht.

A VPT (vibration perception threshold)N25. Valk score N4.

History of rheumatoid arthritis,Severe foot trauma, severe deformity i.e.which requires orthopaedic shoesSurgery of the foot.ABPI

Gomes et al. (2011) Volunteers Michigan neuropathy screeninginstrument questionnaire score of N3/13.Michigan Neuropathy Screening Formscore of N4/10 N5 years type 2 diabetesduration.

N65 years of age,Partial or total foot amputation,Charcot arthropathy,Orthopaedic foot alteration confirmed byradiography,Any additional central or peripheral nervedisorder,Presence of retinopathy, nephropathy orplantar ulcersInability to walk.

Caselli et al. (2002) Participants from a large multicenterprospective study,

Neuropathy disability score was used(NDS). Both motor and sensorycomponents measured.

Not defined

Bacarin et al. (2009) Cases recruited through The NationalAssociation for Assistance to Diabetics. Thecontrols were employee volunteers fromthe Academic department that conductedthis study.

N5 year diabetes durationN6 on the Michigan NeuropathyquestionnaireAt least 2 areas where a 1 Ogmonofilament was unfelt.

N65 years of age,Hallux amputation or partial amputationof the foot, major foot shape alterations byvisual inspection,Orthopaedic disorders of the lower limbsor pain, or use of any assistive devices forwalking (walking sticks/canes),Traumatic ulceration that could beimmediately recognized by the subject.Charcot arthropathy confirmed byradiography or foot ulcers.

836 M. Fernando et al. / Clinical Biomechanics 28 (2013) 831–845

Table 3 (continued)

Study Recruitment source Neuropathy classification for inclusion Exclusion criteria

Yavuzer et al. (2006) Outpatient diabetes follow-up clinic of auniversity hospital

Muscle EMG N2 standard deviations fromnormal values.Reduced peripheral nerve conduction ve-locity (peroneal and tibial motor nerves).

Foot ulcers, amputation, Charcot's joints,neurological, musculoskeletal orrheumatic disease.

Uccioli et al. (2001) Outpatient clinic Neuropathy disability score (NDS) N 6/10,Vibration perception threshold N25 andcontralateral limb N20 measured with aneurothesiometer.

Foot ulcers,N65 yearsInability to walkNeurological, musculoskeletal orrheumatological conditions, PVD,Claudication pain, ABPI b 0.85, Charcotjoints, amputation.

Legend: Characteristics of participant recruitment sites, assessment of neuropathy and exclusion criteria used according to study.Caselli et al. (2002) reported findings for three neuropathy groups. The severe neuropathy group was used in comparison with the diabetes only group in this review.

a Applies to Savelberg et al. (2010) only.

837M. Fernando et al. / Clinical Biomechanics 28 (2013) 831–845

length. Therewas also a high level of heterogeneity present. Two studiesreported stride length findings for the DPN group compared to the HCgroup and both studies reported lower values in DPN patients(Savelberg et al., 2010; Sawacha et al., 2009b).

One study reported stance phase duration as a percentage of the gaitcycle (Sawacha et al., 2012b). These findings were consistent with theabove findings suggesting that DPN patients had longer percentage du-ration in the stance phase of gait (Table 4). Meta-analysis combiningdata from three studies (DPN n = 54, DMC n = 51) suggested that pa-tients with DPN had a longer stance time at a moderate effect level(standardised mean difference 0.54, 95% CI 0.15–0.93; P = 0.006). Theheterogeneity between studies was minimal I2 = 0.

3.5. Kinematics

Only one study reported on kinematics at the hip, knee and ankle inboth extension and flexion directions (Table 5) (Gomes et al., 2011).While DPN participants exhibited greater hip flexion (degrees) whencompared to HC subjects, the DPN participants also demonstrated re-duced hip extension, knee flexion and knee extension when comparedto the HC group. Both maximum ankle plantar flexion and ankledorsiflexion were reduced in DPN participants when compared to theHC group. Meta-analysis was not possible for these results.

3.6. Kinetics

Five studies reported kinetic variables (Saura et al., 2010; Savelberget al., 2009; Sawacha et al., 2012a; Uccioli et al., 2001; Yavuzer et al.,2006) (Tables 5 and 6). Two studies reported on the force generationcomponents at the ankle, knee and hip (Savelberg et al., 2009;

Table 4Temporal–spatial gait parameters.

Study Walking speed (m/s) Stride length (m)

DPN DMC HC DPN DMC

Sawacha et al. (2009a) 1.10(0.2)

1.10(0.2)

1.27(0.1)

1.20(0.2)

1.23(0.1)

Sawacha et al. (2009b) 1.00 (0.2) 1.00 (0.1)Savelberg et al. (2009) 1.40

(0.09)1.30(0.13)

1.22(0.20)

Sawacha et al. (2012b) 1.11(0.21)

1.23(0.21)

1.24(0.19)

1.33(0.2)

Gomes et al. (2011) 0.9(0.4)

1.03(0.4)

Savelberg et al. (2010) 1.02(0.13)

1.06(0.13)

1.18(0.22)

1.15(0.15)

1.22(0.13

Yavuzer et al. (2006) 0.94(0.3)

0.85(0.2)

Mean and (SD) diabetic neuropathy subjects (DPN), diabetes subjects (DMC), healthy controls

Yavuzer et al., 2006) (Table 6). According to one study, both the brakingand propelling forces were reduced in the DPN group compared to bothDMC and HC groups (Savelberg et al., 2009). Both the first maximumsupport moment and the mid stance minimal support momentwere elevated in DPN participants compared to the DMC and HCgroups; however the second maximum support moment was slight-ly higher in the DMC group when compared to DPN patients(Sawacha et al., 2009b).

The results for maximum ankle plantar flexionmomentwere incon-sistent. One study reported a higher value in DPN patients compared tocontrols (Sawacha et al., 2009b); while another study reported a lowervalue in DPN patients when compared to HC subjects (Yavuzer et al.,2006). Results for knee extension moments were also inconsistent.One study reported reduced extension moments in DPN patients(Sawacha et al., 2009b) and another higher extension moments inDPN patients when compared to both the DMC and HC groups(Yavuzer et al., 2006). However, both studies reported greater kneeflexion moment in the DPN group compared to both the DMC and HCgroups (Savelberg et al., 2009; Yavuzer et al., 2006) (Table 6).

According to a single study, the hip extension moment was greaterin the DPN group when compared to both control groups (Sawachaet al., 2009b). According to both studies, the hip flexion moment wasalso reduced in DPN patients compared to both controls (Sawachaet al., 2009b; Yavuzer et al., 2006).

Meta-analysis was only possible for the vertical GRF (first peak) atinitial contact. Although reported vertical GRF was higher in DPN pa-tients compared to both HC and DMC groups, the results from themeta-analysis were statistically insignificantwith a high level of hetero-geneity. Meta-analysis was not possible for vertical GRF at toe-off(second peak) (Table 5). However, one study reported a higher vertical

Stance time (s) Stance phase (% gait cycle)

HC DPN DMC HC DPN DMC HC

1.4(0.1)

0.73(0.1)

0.66(0.1)

0.5(0.1)

0.7(0.10)

0.64(0.0)

61.07(3.14)

59.7(2.20)

59.4(2.32)

)1.28(0.15)

0.64(0.5)

0.62(0.4)

0.61(0.3)

(HC). Empty boxes represent missing data or unreported data.

Table 5Ground reaction forces and joint kinematics.

Study

Go

me

s(2

011

)S

awa

cha

(20

12

a)

Ucc

ioli

(20

01

)Y

avu

zer

(20

06

)S

au

ra(2

01

0)

Vertical GRF

initial cntact

(N/ Body weight %)

DP

N

DM

C

HC

DP

N

DM

C

HC

DP

N

DM

C

HC

DP

N

DM

C

HC

DP

N

DM

C

HC

DP

N

DM

C

HC

DP

N

DM

C

HC

DP

N

DM

C

HC

Vertical GRF at

toe-off

(N/ Body weight %)

Maximum hip

flexion

(deg)

Maximumhip

extension

(deg)

Maximum knee

flexion

(deg)

Maximum knee

extension

(deg)

Maximum Ankle

plantar flexion

(deg)

Maximum Ankle

dorsiflexion

(deg)

20.20

(3.86)

82.50

(2.87)

87.3

(8.4)

95.0

(5.3)

103.88

(4.82)

80.64

(3.17)

93.8

(8.4)

94.2

(5.6)

91.20

(4.4)

95.6

(4.4)

106.38

(8.3)

95.0

(4.3)

93.6

(6.9)

96.5

(5.4)

93.82

(5.3)

91.3

(9.0)

93.1

(4.9)

91.80

(8.5)

17.97

(3.46)

5.69

(1.73)

6.06

(1.95)

26.30

(4.84)

29.50

(5.49)

26.41

(19.77)

28.53

(16.31)

9.01

(3.27)

11.79

(5.74)

5.63

(2.48)

5.80

(1.72)

Barefoot gait findings mean and (SD).All joint range of motion (flexion/extension) reported for stance phase duration. Empty boxes represent missing data or unreported data.

838 M. Fernando et al. / Clinical Biomechanics 28 (2013) 831–845

GRF value in DPN patients at toe-off (Saura et al., 2010) and another alower value (Yavuzer et al., 2006).

3.7. Dynamic EMG

Muscle activationwas reported for several different lower limbmus-cle groups (Table 7). Two studies reported findings as % stance phase(Akashi et al., 2008; Gomes et al., 2011) and one study as % gait cycle(Sawacha et al., 2012b). Three studies reported the duration of activityof the tibialis anterior muscle (Akashi et al., 2008; Gomes et al., 2011;Sawacha et al., 2012b). Meta-analysis suggested a non-significant lon-ger duration of muscle activity in the tibialis anterior muscle in DPN pa-tients when compared to the HC group.

Meta-analysis was not possible for the other muscle groups due tolack of studies. However, according to two studies, the lateral gastrocne-mius muscle had reduced duration of activity in DPN patients (% stance

Table 6Joint moments and forces during gait.

Study Braking force(N/Kg)

Propellingforce (N/Kg)

Max ankle plantarflexion moment(Nm/Kg)

Max kneeextensionmoment (Nm/Kg

Max kneeflexion moment(Nm/Kg

DP

N

DM

C

HC

DP

N

DM

C

HC

DP

N

DM

C

HC

DP

N

DM

C

HC

DP

N

DM

C

HC

Sav

elb

erg

(20

09

*)Y

avu

zer

(20

06

)

1.3

0 (

0.3

1)

1.6

9 (

0 .2

9)

1.74

(0

.45

)

1.4

2 (

0.3

5)

2.0

2 (

0.3

4)

1.7

8 (

0.5

)

1.6

4 (

0 .2

6)

1.2

(0

.2)

1.5

1 (

0.2

11.

1 (

0.3

)

1.5

9 (

0.1

07

1.3

(0

.2)

0.2

3 (

0.3

0)

0.8

1 (

0.5

6)

0.4

2 (

0.2

2)

0.5

2 (

0.5

6)

0.5

2 (

0.5

6)

0.4

48

(0

.63

)

0.2

9 (

0.2

1)

0.8

1 (

0.1

8)

0.1

6(0

.11

)0

.43

(0

.12

)

0.4

7 (

0.1

9)

0.2

8 (

0.2

7)

Mean and (SD) diabetic neuropathy subjects (DPN), diabetes subjects (DMC) healthy controls⁎Study reported values which were normalised to individual body weight of participants.Empty boxes represent missing data or unreported data.

phase) when compared to the HC group (Akashi et al., 2008; Sawachaet al., 2012b). On the contrary, assessment of the vastus lateralis musclesuggested a longer duration of activation in DPN patients when com-pared to the HC group (Akashi et al., 2008; Gomes et al., 2011).

There were conflicting results from two studies which assessed ac-tivity of the peroneus longus muscle (Table 7). One study reported re-duced duration of muscle activation in DPN patients (% stance phase)compared to the HC group; and another study reported longer duration(% gait cycle) in DPN patients compared to both HC and DMC groups.

The findings from the assessment of the gluteusmediusmuscle, rec-tus femorismuscle andmedial gastrocnemiusmuscleswere from singlestudies and are highlighted in Table 7. Both gluteus medius and rectusfemoris muscles were reported to have reduced duration of activity(Sawacha et al., 2012b), while the medial gastrocnemius was reportedto have a longer duration of activity (Gomes et al., 2011) in DPNpatients.

Max Hipextension moment (Nm/Kg

Max Hip extension moment (Nm/Kg

First maxsupport moment (Nm/Kg

Midstance minsupport moment (Nm/Kg

Second maxsupport moment (Nm/Kg

DP

N

DM

C

HC

DP

N

DM

C

DP

N

DM

C

HC

HC

DP

N

DM

C

HC

DP

N

DM

C

HC

1.0

8 (

0.3

9

1.0

9 (

0.2

5)

0.8

5 (

0.4

0)

0.7

5 (

0.3

6)

0.5

5 (

0.2

6)

0.7

0 (

0. 1

2)

0.4

9 (

0.3

2)

0.5

0 (

0.3

9)

0.5

1 (

0.2

0)

1.8

7 (

0.3

2)

1.8

4 (

0.3

3)

1.5

2 (

0.4

7)

1.8

7 (

0.3

2)

0.8

5 (

0.3

1

0.6

(0

.21

1.18

(0

.33

1.2

0 (

0.4

4)

0.8

6 (

0.2

0)

(HC) of joint moments and force during gait.

Table7

Electrom

yograp

hy(EMG)pe

akactiva

tion

times

during

stan

ce.

Stud

yRe

ctus

femoris

(%of

supp

ort)

Tibialisan

terior

(%of

supp

ort)

Gluteus

med

ius

(%of

supp

ort)

Lateralg

astrocne

mius

(%of

supp

ort)

Med

ialg

astrocne

mius

(%of

supp

ort)

Vastuslateralis

(%of

supp

ort)

Perone

uslong

us(%

ofsupp

ort)

DPN

DMC

HC

DPN

DMC

HC

DPN

DMC

HC

DPN

DMC

HC

DPN

DMC

HC

DPN

DMC

HC

DPN

DMC

HC

Sawacha

etal.(20

12b)

5.46

(1.26)

a6.82

(1.24)

a11

.8(1.29)

a11

.71

(1.13a

6.96

(1.10)

a9.27

(1.63)

a7.55

(2.68)

a11

.7(1.87)

a13

.2(1.89)

a38

.1(1.66)

b35

.9(1.38)

b41

.60

(2.29)

b41

.81

(0.09)

b37

.2(0.11)

b33

.4(0.21)

b

Aka

shie

tal.(20

08)

6.10

(1.68)

6.05

(2.15)

62.84

(5.06)

63.53

(3.65)

11.97

(2.31)

10.82

(3.33)

Gom

eset

al.(20

11)

3.42

(1.73)

4.33

(1.80)

60.16

(6.99)

54.33

(6.18)

10.37

(3.18)

9.02

(3.90)

56.09

(11.22

)60

.9(7.98)

Barefoot

dyna

micEM

Gfind

ings

during

gaitas

timeof

activa

tion

peak

aspe

r%of

supp

ortph

asemeanan

d(SD)bo

thco

mbine

dleftan

drigh

tmeasu

remen

tun

less

othe

rwisesp

ecified

.Note—

Sawacha

etal.(20

12b)

repo

rted

find

ings

as%ga

itcycle.

Emptybo

xesrepresen

tmissing

data

orun

repo

rted

data.

aFind

ings

wererepo

rted

forinitialc

ontact

toload

ingrespon

se.

bFind

ings

wererepo

rted

formid-stanc

e.

839M. Fernando et al. / Clinical Biomechanics 28 (2013) 831–845

3.8. Plantar pressure (peak pressure and pressure time integral)

Six studies reported plantar pressure data of interest (Bacarin et al.,2009; Caselli et al., 2002; Guldemond et al., 2008; Melai et al., 2011;Sacco et al., 2009; Sawacha et al., 2012a) (Table 8). Themajority of stud-ies reported plantar pressure as MPP while three studies reported PTI(Bacarin et al., 2009; Melai et al., 2011; Sacco et al., 2009).

Meta-analysis combining data from three studies (DPN n = 108,HC n = 55) suggested that patients with DPN had elevated plantarpressure (both MPP and PTI) at the rearfoot at moderate effect levels(MPP standardised mean difference 0.45, 95% CI 0.09–0.82 P ≤ 0.001,I2 = 7.0; and PTI standardised mean difference 0.40, 95% CI 0.05–0.75P = 0.02 I2 = 0). Both results contained minimal heterogeneity.Meta-analysis results for MPP at the rearfoot were insignificant forDPN patients when compared to DMC patients (Table 9).

Meta-analysis results for the midfoot (DPN n = 108, HC n = 55,combining three studies) revealed greater MPP and PTI in DPN patients(MPP standardised mean difference 0.72, 95% CI 0.37–1.08 P ≤ 0.001I2 = 0; and PTI standardised mean difference 0.50, 95% CI 0.15–0.85P = 0.005 I2 = 7.0). There was minimal heterogeneity betweenstudies.

Meta-analysis for plantar pressure at the forefoot (DPN n = 177,DMC n = 102, HC n = 55, combining three studies) demonstratedgreater MPP in the forefoot of DPN patients at moderate effect levelscompared to the HC group (standardised mean difference 0.55, 95% CI0.20–0.90 P = 0.002 I2 = 0) and DMC group (standardised meandifference 0.51, 95% CI 0.24–0.78 P ≤ 0.001 I2 = 10.1) respectively.Furthermore, meta-analysis for PTI at the forefoot (DPN n = 177, HCn = 55, combining three studies) suggested that forefoot PTI was alsoelevated in DPN patients at moderate effect levels (standardised meandifference 0.66, 95% CI 0.31–1.02; P ≤ 0.001; I2 = 0). There was mini-mal heterogeneity between studies. Meta-analysis results for the hallux(MPP and PTI) comparing plantar pressure between the three groupsrevealed non-significant differences (Table 9).

Findings from two studies suggested that MPP at the plantar aspectof thefirstmetatarsophalangeal jointwas higher in theDPN group com-pared to the DMC group (Guldemond et al., 2008; Melai et al., 2011),while results from one study suggested that MPP at the plantar aspectof the first metatarsophalangeal joint was higher in DPN patients com-pared to the HC group (Melai et al., 2011). According to a single study,the PTI values were higher in DPN patients compared to both DMCand HC groups (Melai et al., 2011). However, according to the samestudy, there was a lower PTI and MPP for the DPN group in the lessertoes compared to both control groups (Melai et al., 2011) (Table 7).

4. Discussion

To the best of the authors' knowledge this is the first systematic re-view and meta-analysis of studies investigating the gait cycle, muscleactivation and plantar pressure exclusively in DPN patients comparedto DMC and HC groups. The aim of this review and meta-analysis wasto assess the gait dissimilarities between DPN, DMC and HC subjects inrelation to spatiotemporal, kinetic, kinematic, EMG and plantar pressurevariables. Ourfindings,within the limitations of the review, indicate gaitdifferences in DPN patients when comparedwith DMC and HC subjects,likely resulting from sensory and motor neuropathy (Andersen, 2012;Kovac et al., 2011). The primary advantage of relating both HC andDMC groups to DPN patients was the ability to appreciate subtle differ-ences within each group for comparison and contrast. However, it mustbe emphasised that therewas a high level of heterogeneity formost var-iables between studies as highlighted by the Q and I2 statistics. This highlevel of heterogeneitywas also evident in other systematic reviewers in-vestigating plantar pressures in similar patient groups (Crawford et al.,2007; Monteiro-Soares et al., 2012).

DPN is a significant complication of diabetes and accounts for signif-icant morbidity and mortality (Boulton, 1998; Cook and Simonson,

Table 8Peak plantar pressure and pressure time integral.

Study Rear-foot Mid-foot Metatarsal 1 Fore-foot Hallux Lesser toes

DPN DMC HC DPN DMC HC DPN DMC HC DPN DMC HC DPN DMC HC DPN DMC HC

Melai et al. (2011)a

MPP (N/cm2) 42.5(11.8)

41.9(10.9)

35.9(9.3)

15.0(5.2)

16.5(6.0)

11.8(2.4)

44.9(24.4)

31.5(12.1)

27.3(9.2)

50.1(19.8)

44.8(13.3)

36.4(7.5)

46.3(24.3)

51.4(28.6)

35.5(14.9)

15.3(7.1)

15.4(8.1)

16.2(7.2)

PTI (Ns/cm2) 3.4(0.9)

3.5(0.7)

3.1(0.7)

1.2(0.6)

1.5(0.6)

1.0(0.1)

4.7(2.0)

3.9(1.0)

3.2(1.0)

5.9(1.3)

5.8(1.4)

5.0(1.0)

2.6(1.1)

2.7(1.1)

2.5(1.0)

1.0(0.5)

1.0(0.6)

1.3(0.7)

Guldemond et al. (2008)b

MPP (N/cm2) 48.1(31.3)

30.8(13.8)

68.9(27.9)

55.1(22.6)

40.5(25.7)

45.5(26.4)

Bacarin et al. (2009)c

MPP (N/cm2) 34.2(7.6

33.7(9.5)

20.5(11.8)

13.9(7.6)

36.5(9.3)

34.7(8.8)

30.5(11.1)

30.6(11.0)

PTI (Ns/cm2) 9.4(2.9)

8.3(2.1)

4.3(0.9)

3.7(1.1)

11.0(2.6)

9.7(2.3)

7.4(2.6)

6.8(2.4)

Sacco et al. (2009)d

MPP (N/cm2) 22.0(4.0)

19.6(2.7)

11.4(5.2)

7.5(3.1)

24.5(5.6)

21.8(3.5)

PTI (Ns/cm2) 2.7(0.5)

2.5(0.3)

3.9(1.7)

3.0(0.9)

5.3(1.6)

4.4(0.8)

Sawacha et al. (2012a)b

MPP (N/cm2) 77.5 42.7 51.5 31.2 41.0 58.4

Caselli et al. (2002)b

MPP (N/cm2) 23.0(10.0)

32.0(20.0)

62.0(40.5)

33.0(21.0

Reported right foot plantar pressures and pressure time integrals normalised to body weight (mean) and (SD).a This study reported multiple forefoot plantar pressures; the highest plantar pressure data-set at metatarsal 3 was taken as the forefoot plantar pressure measurements.b These studies did not report PTI.c This study reported both medial and lateral plantar pressures for the forefoot. The highest valued pair (medial) was extrapolated for comparison.d This study reported plantar pressure at initial strike and at toe-off, the initial strike pressures were utilised for rear foot and the toe off pressures for the forefoot and mid-foot as these were the highest pressures reported.

840M.Fernando

etal./ClinicalBiomechanics

28(2013)

831–845

Table 9Meta-analyses results.

Outcome measure Comparison Number of studies Effect size (95%-CI) P value Heterogeneity assessment Classic fail safe N

Walking speed [m/s] DPN versus DMC 5 −0.041(−0.563, 0.482)

P = 0.879 Q = 10.7; P = 0.030; I2 = 62.5 0

DPN versus HC 5 −0.270(−0.936, 0.396)

P = 0.427 Q = 14.6; P = 0.006; I2 = 72.6 0

Stride length DPN versus DMC 3 −0.343(−0.730, 0.044)

P = 0.083 Q = 0.54; P = 0.763; I2 = 0 0

Stance time DPN versus DMC 3 0.545(0.153,0.937

P = 0.006 Q = 1.4; P = 0.496; t2 = n 2

Vertical GRF initial contact DPN versus DMC 3 0.543(−0.615, 1.701)

P = 0.358 0 = 17.0; P b 0.001; I2 = 0 0

DPN versus HC 4 0.614(−0.601, 1.828)

P = 0.322 Q = 30.6; P b 0.001; I2 = 90.2 0

Vertical GRF toe off DPN versus DMC 3 0.392(−0.647, 1.430)

P = 0.460 0 = 14.1; P = 0.001; I2 = 85.8 0

DPN versus HC 3 0.142(−1.244, 1.529)

P = 0.841 Q = 22.2; P b 0.001; I2 = 91.0 0

EMG tibialis anterior muscle DPN versus HC 3 0.416(−0.841, 1.674)

P = 0.516 0 = 19.1; P b 0.001; I2 = 89.6 0

MPP — rear foot DPN versus HC 3 0.455(0.090, 0.821)

P = 0.015 Q = 2.2; P = 0.341; I2 = 7.0 2

PTI — rear foot DPN versus HC 3 0.407(0.058, 0.756)

P = 0.022 0 = 0.115; P = 0.944; I2 = 0 2

MPP — mid foot DPN versus HC 3 0.729(0.373, 1.084)

P b 0.001 Q = 0.33; P = 0.848; I2 = 0 10

PTI — mid foot DPN versus HC 3 0.503(0.152, 0.854)

P = 0.005 Q = 0.54; P = 0.765; I2 = 0 4

MPP — forefoot DPN versus DMC 3 0.510(0.240, 0.780)

P b 0.001 Q = 2.23; P = 0.329; I2 = 10.1 10

DPN versus HC 3 0.552(0.200, 0.903)

P = 0.002 0 = 1.8; P = 0.415; I2 = 0 4

PTI — forefoot DPN versus HC 3 0.666(0.313, 1.020)

P b 0.001 Q = 0.22; P = 0.894; I2 = 0 8

Meta-analyses of studies comparing healthy controls (HC) with control patients with diabetes mellitus (DMC) with patients with diabetic peripheral neuropathy (DPN).Effect size is standardised difference ofmean values calculated as DPN— (control group); hence a positive result implies larger value for DPN and negative values reduced values for DPN.Results are from random effect models. Forest Plots for the significant results can be found in additional figures.

841M. Fernando et al. / Clinical Biomechanics 28 (2013) 831–845

2012). The primary risk factor for DPN is hyperglycaemia as it leads toincreased oxidative stress, production of advanced glycation end prod-ucts, increased polyol pathway flux and protein kinase C activation. Allthese factors are believed to contribute to micro-vascular disease andnerve dysfunction (Park et al., 2004). The end result of DPN can be cat-astrophic for patients, as this leads to foot ulceration and increased riskof limb amputation, significant healthcare costs, reduced quality of lifeand reduced mobility (Boulton, 2005; Price, 2004; Singh et al., 2005).Therefore, understanding the impact of DPN on the biomechanical as-pects of human locomotion is clinically important (Formosa et al.,2013).

We hypothesised that spatiotemporal parameters would be signifi-cantly reduced in DPN patients compared to both controls. Themajorityof reported findings indicated that DPN patients walked slower and hadreduced stride length when compared to both DMC and HC groups,however, meta-analysis results were statistically insignificant. Theonly significant finding was that DPN patients expended a longerperiod of time in the stance phase compared to DMC subjects. Wehypothesised that the force generation at the hip, knee and ankle

Fig. 2. DPN versus DMC stride le

would be significantly increased for both flexion and extension mo-ments in participants with DPN. There were insufficient studies tocarry out meta-analysis and the two studies which reported findingsdemonstrated conflicting results. Regardless of the fact that one studyutilised a significantly younger HC group, the differences between stud-ies could not be solely explained by a difference in the age groups(Yavuzer et al., 2006). Irrespective of this, increased knee flexion mo-ment in the DPN group was reported by both studies, emphasisingthat greater force generation may occur during knee flexion in DPNpatients. This finding suggests the possibility that knee flexionmight be an important compensation strategy in patients withDPN, as the motor component of DPN manifests in a stocking andglove distribution and affects the distal joints first (Tesfaye andSelvarajah, 2012).

The first maximum support moment (combination of extensor mo-ments at hip, knee and ankle) (Winter, 1980) was higher in the DPNgroup when compared to the DMC and HC groups (Savelberg et al.,2009). Although reported in a single study, this suggests that combinedforces at the hip; knee and ankle during the stance phase are greater in

ngth (Table 3) Forest Plot.

Fig. 3. DPN versus HC; peak plantar pressure rearfoot (Table 7) Forest Plot.

Fig. 4. DPN versus HC; pressure time integral rearfoot (Table 7) Forest Plot.

842 M. Fernando et al. / Clinical Biomechanics 28 (2013) 831–845

DPN patients compared to both control subjects (Savelberg et al., 2009).Further studies are needed to confirm this finding.

Even though meta-analysis of the vertical GRF demonstrated thatDPN patients had a higher initial contact force than DMC and HC sub-jects, the level of heterogeneity in studies was high and the meta-analysis results were statistically insignificant. It was anticipated thatDPN patients would exhibit higher GRF due to neurological deficit andreduced proprioception but the current findings fail to support this hy-pothesis. Similarly, we hypothesised that participants with DPN wouldexhibit reductions in joint RoM at the hip, knee and ankle during gait,as a result ofmotor neuropathy (Andersen, 2012). Therewere few stud-ies investigating lower limb kinematics of DPN patients during locomo-tion to investigate this hypothesis. One study (Sawacha et al., 2012a)reported kinematic variables of the foot which were outside the scopeof this review and were not included. Therefore, current findings forjoint angle kinematics were drawn from one publication investigatingbarefoot lower limb kinematics (Gomes et al., 2011). With the excep-tion of hip flexion, the findings demonstrated reduced RoM in DPN pa-tients compared to HC subjects. This finding was consistent with ourhypothesis. A higher proportion of hip flexion is also another possiblecompensatory mechanism to increase stability in the gait strategy ofDPN patients. Increased hip flexion could also be a compensatorymech-anism to adjust for impaired ankle dorsiflexion in patients with DPN.We did not directly examine this possibility in the current review. Fur-ther studies are needed to clarify the cause of greater joint force in kneeflexion and greater degree of hip flexion in patients that have DPN.

Dynamic EMG data suggested that the tibialis anterior muscleremained active for a longer duration of time in DPN patients comparedto HC subjects. Meta-analysis suggested that this findingwas not statis-tically significant and demonstrated a high level of heterogeneity.Therefore, it was difficult to ascertain whether this was consistentwith our hypothesis of altered muscle activity duration in DPN

Fig. 5. DPN versus HC; peak plantar pre

participants due to mis-firing and reduction in neural pathways associ-ated with muscle recruitment and deactivation. It was also challengingto explain the shorter duration of activity of the lateral gastrocnemiusmuscle and longer duration of activity of the vastus lateralis muscleand the various reported findings of other muscle groups from individ-ual studies. It seems that there are clear differences in muscle activationbetweenDPN, DM andHC subjects; however thefindings frompreviousstudies were not consistent. It could be possible that these observationswere due to changes in action potential amplitude and inconsistency inthe number ofmotor units recruited during EMGmeasurement of lowerlimb muscle activation in DPN patients, however there is currently in-sufficient data to support this theory. As hypothesised, the meta-analysis results suggested that DPN participants have higher dynamicplantar pressures at rearfoot,midfoot and forefoot siteswhen comparedto controls. However, there were insufficient studies to carry out ameta-analysis of data collected at the hallux and lesser toe joints andthe results from studies were highly contradictory.

Previous reviews have highlighted gait differences in patients thathave diabetes mellitus, but have not concentrated on DPN as the mainfocus (Allet et al., 2008; Wrobel and Najafi, 2010). The limitations ofthis review were the small number of included studies, the small num-ber of participants in included studies, the high level of heterogeneitybetween studies, the investigation of barefoot measurements only, theexclusion of kinematic data of the foot and the language limitation tostudies written in English.

We can conclude from the current level of evidence that the onlybiomechanical factors that seem significantly different in DPN patientscompared to DMC and HC groups are elevated plantar pressureand longer stance time, illustrated by moderate effect sizes fromstandardised mean differences. Therefore, it is probable that elevatedplantar pressure coupled with a longer period of time spent in stancein DPN patients contributes to the susceptibility for skin damage

ssure midfoot (Table 7) Forest Plot.

Fig. 7. DPN versus DMC; peak plantar pressure forefoot (Table 7) Forest Plot.

Fig. 6. DPN versus HC; pressure time integral midfoot (Table 7) Forest Plot.

843M. Fernando et al. / Clinical Biomechanics 28 (2013) 831–845

through prolonged mechanical load on tissue, leading to skin break-down and ulceration (van Dieren et al., 2010). Although it is possiblethat reduced spatiotemporal parameters, elevated vertical GRF, longermuscle duration and reduced joint kinematics contribute to foot ulcera-tion; the current knowledge base is insufficient for firm conclusions.There were significant discrepancies between studies reporting find-ings. Our observations were similar to that of Allet et al. and Wrobeland colleagues (Allet et al., 2008; Wrobel and Najafi, 2010).

While all studies in this review utilised procedures for diagnosingDPN in participants, only two studies excluded patients with PAD. PADhas been reported to have significant effects on walking patterns(Crowther et al., 2007, 2008). The BMI of all three groups was similarand it is unlikely that this accounted for any difference in gait variables.The mean diabetes duration between DPN and DMC subjects was notsignificantly different in the studies included. It has been hypothesisedthat DPN can manifest in people with a diabetes duration greater than10 years, as it does in those with poor glycaemic control (Kovac et al.,2011; Oguejiofor et al., 2010; Valensi et al., 1997). In addition, smallfoot muscle atrophy resulting from the effects of hyperglycaemia andsmall nerve damage have also been confirmed in diabetes patientsutilising MRI, before DPN becomes clinically detectable (Greenmanet al., 2005). Therefore, these factors may also influence gait findingsinDMCgroupswhen compared to DPNpatients. This could be a possibleexplanation for the similar results in DPN and DMC subjects and lack ofstatistical significance. However, the scope of this review was also de-pendent on the sample sizes of original studies, and thus, reported sta-tistical insignificant differences may have been due to lack of power.

There is paucity in biomechanical literature investigating the effectsof DPN on barefoot gait parameters, particularly in relationship to theeffects of severe neuropathy resulting in foot lesions and its effect onhuman locomotion. The clinical ramifications from this systematic re-view are limited due to the high level of heterogeneity and statistically

Fig. 8. DPN versus HC; peak plantar pre

insignificant results from the meta-analyses. However, it was evidentthat patients with DPN demonstrated greater overall dynamic plantarpressure and forefoot plantar pressure (both MPP and PTI) comparedto patients without DPN. Patients with DPN also expended a longer du-ration of time in the stance phase. Both findings potentially contributetowards ulceration in patients with DPN. Other biomechanical findingswere less clear and we therefore encourage future biomechanical stud-ies inDPN to assess factors such as lower limbangular kinematics, kinet-ics and EMG and to adjust for variables such as PAD, claudication painand history of foot ulcers in selection of participants with DPN asthese factors are highly likely to influence walking patterns.

5. Conclusion

Current evidence from the literature indicates that DPN patients ex-hibit significantly elevated plantar pressures and occupy a longer dura-tion of time in stance phase during gait compared to controls. Weencourage future biomechanical studies in DPN to assess factors suchas lower limb angular kinematics, kinetics and EMG.

Author details

Mr. Malindu (Mal) Fernando BHscPod (Hons), Clinical Podiatrist andCohort Doctoral PhD Candidate James Cook University, Townsville.

Dr. Robert Crowther BScExSc (Hons), PhD, Exercise Physiologist, Lec-turer, Manager — Movement Analysis Laboratory, James Cook UniversityTownsville.

Mr. Peter Lazzarini BPodiatry, Senior Research Fellow (Podiatry) AlliedHealth Research Collaborative, Metro North Hospital and Health Service,Brisbane And School of Clinical Sciences, Queensland University ofTechnology

ssure forefoot (Table 7) Forest Plot.

Fig. 9. DPN versus HC; pressure time integral forefoot (Table 7) Forest Plot.

844 M. Fernando et al. / Clinical Biomechanics 28 (2013) 831–845

Dr. Kunwarjit Sangla, MBBS, MD, FRACP, Consultant Endocrinologist,Director of Internal Medicine, Townsville Hospital.

Dr. Margaret Cunningham, PhD, Health Psychologist and Clinical Re-search Team Leader Vascular Biology Unit, James Cook University.

Associate Prof. Petra Buttner, PhD Biostatistician and Epidemiologist,School of Public Health and Tropical Medicine, James Cook University.

Prof. Jonathan Golledge, MB BChir, MA, MChir, FRCS, FRACS, DirectorDepartment of Vascular Surgery Townsville Hospital, Head of QueenslandResearch Centre for Peripheral Vascular Disease, James Cook University,Townsville.

Conflicts of interest

The authors wish to declare no relevant conflicts of interest.

Acknowledgements

Funding from the Commonwealth government, Queensland Gov-ernment and National Health and Medical Research Council supportedthis work. JG holds a Practitioner Fellowship from the National Healthand Medical Research Council, Australia (1019921). JG holds a SeniorClinical Research Fellowship from the Office of Health and MedicalResearch. MF is currently supported by an Australian PostgraduateAward Scholarship at James Cook University.

References

Akashi, P.M., Sacco, I.C., Watari, R., Hennig, E., 2008. The effect of diabetic neuropathyand previous foot ulceration in EMG and ground reaction forces during gait. Clin.Biomech. 23, 584–592.

Allet, l, Armand, S., Golay, A., Monnin, D., De Bie, R., De Bruin, E., 2008. Gait characteristicsof diabetic patients: a systematic review. Diabetes Metab. Res. Rev. 24, 173–191.

Andersen, H., 2012. Motor dysfunction in diabetes. Diabetes Metab. Res. Rev. 28(Suppl. 1), 89–92.

Armstrong, D.G., Lavery, l.A., Holtz-Neiderer, K., Mohler, M.J., Wendel, C.S., Nixon, B.P., etal., 2004. Variability in activity may precede diabetic foot ulceration. Diabetes Care 27,1980–1984.

Bacarin, T.A., Sacco, I.C., Hennig, E.M., 2009. Plantar pressure distribution patterns duringgait in diabetic neuropathy patients with a history of foot ulcers. Clinics (Sao Paulo)64, 113–120.

Boulton, A.J., 1998. Guidelines for diagnosis and outpatient management of diabetic pe-ripheral neuropathy. European Association for the Study of Diabetes, Neurodiab. Dia-betes Metab. 24 (Suppl. 3), 55–65.

Boulton, A.J.M., 2005. Management of diabetic peripheral neuropathy. Clin. Diab. 23, 9–15.Boulton, A.J., Vileikyte, L., Ragnarson-tennvall, G., Apelqvist, I., 2005. The global burden of

diabetic foot disease. Lancet 366, 1719–1724.Caselli, A., Pham, H., Giurini, J.M., Armstrong, D.G., Veves, A., 2002. The forefoot-to-

rearfoot plantar pressure ratio is increased in severe diabetic neuropathy and canpredict foot ulceration. Diabetes Care 25, 1066–1071.

Cavanagh, P., Ulbrecht, J., Caputo, G., 2000. New developments in the biomechanics of thediabetic foot. Diabetes Metab. Res. Rev. 16, s6–s10.

Cohen, J., 1988. Statistical Power Analysis of the Behavioural Sciences. Lawrence ErlbaumAssociates Inc., United States.

Cook, J., Simonson, D., 2012. Epidemiology and Health Care Cost of Diabetic Foot Prob-lems. Springer.

Crawford, F., Inkster, M., Kleijnen, J., Fahey, T., 2007. Predicting foot ulcers in patients withdiabetes: a systematic review and meta-analysis. QJM 100, 65–86.

Crowther, R.G., Spinks, W.L., Leicht, A.S., Quigley, F., Golledge, J., 2007. Relationship be-tween temporal–spatial gait parameters, gait kinematics, walking performance, exer-cise capacity, and physical activity level in peripheral arterial disease. J. Vasc. Surg. 45,1172–1178.

Crowther, R.G., Spinks, W.L., Leicht, A.S., Sangla, K., Quigley, F., Golledge, J., 2008. Effects ofa long-term exercise program on lower limb mobility, physiological responses, walk-ing performance, and physical activity levels in patients with peripheral arterial dis-ease. J. Vasc. Surg. 47, 303–309.

Downs, S.H., Black, N., 1998. The feasibility of creating a checklist for the assessment of themethodological quality both of randomised and non-randomised studies of healthcare interventions. J. Epidemiol. Community Health 52, 377–384.

Formosa, C., Gatt, A., Chockalingam, N., 2013. The importance of clinical biomechanical as-sessment of foot deformity and joint mobility in people living with type-2 diabeteswithin a primary care setting. Prim. Care Diab. 7, 45–50.

Gomes, A.A., Onodera, A.N., Otuzi, M.E.I., Pripas, D., Mezzarane, Rinaldo Andre, Sacco,I.C.N., 2011. Electromyography and kinematic changes of gait cycle at different ca-dences in diabetic neuropathic individuals. Diabet. Neuropathic Gait Muscle Nerve2, 258–268.

Greenman, R.L., Khaodhiar, L., Lima, C., Dinh, T., Giurini, J.M., Veves, A., 2005. Foot smallmuscle atrophy is present before the detection of clinical neuropathy. Diabetes Care28, 1425–1430.

Guldemond, N.A., Leffers, P., Walenkamp, G.H.I.M., Schaper, N.C., Sanders, A.P., Nieman,F.H.M., et al., 2008. Prediction of peak pressure from clinical and radiological mea-surements in patients with diabetes. BMC Endocr. Disord. 8, 16–30.

Kovac, B., Kovac, B., Marusic-Emedi, S., Svalina, S., Demarin, V., 2011. Clinical and electro-physiological signs of diabetic polyneuropathy — effect of glycemia and duration ofdiabetes mellitus. Acta Clin. Croat. 50, 149–157.

Mcgough, J.J., Faraone, S.V., 2009. Estimating the size of treatment effects: moving beyondp values. Psychiatry (Edgmont) 6, 21–29.

Melai, T., Ijzerman, T.H., Schaper, N.C., de lange, T.L.H., Willems, P.J.B., Meijer, K., et al.,2011. Calculation of plantar pressure time integral, an alternative approach. Gait Pos-ture 34, 379–383.

Monteiro-Soares, M., Boyko, E.J., Ribeiro, J., Ribeiro, I., Dinis-Ribeiro, M., 2012. Predic-tive factors for diabetic foot ulceration: a systematic review. Diabetes Metab.Res. Rev. 28 (7), 574–600. http://dx.doi.org/10.1002/dmrr.2319.

Oguejiofor, O.C., Odenigbo, C.U., Oguejiofor, C.B., 2010. Evaluation of the effect of durationof diabetes mellitus on peripheral neuropathy using the United Kingdom screeningtest scoring system, bio-thesiometry and aesthesiometry. Niger. J. Clin. Pract. 13,240–247.

Park, T.S., Park, J.H., Baek, H.S., 2004. Can diabetic neuropathy be prevented? Diab. Res.Clin. Pract. 66, s53–s56 (supplement).

Persaud, R., 1996. Misleading meta-analysis. “Fail Safe N” is a useful mathematical mea-sure of the stability of results. BMJ 312, 125.

Price, P., 2004. The diabetic foot: quality of life. Clin. Infect. Dis. 39, s129–s131.Sacco, I.C.N., Hamamoto, A.N., Gomes, A.A., Onodera, A.N., Hirata, R.P., Hennig, E.M., 2009.

Role of anklemobility in foot rollover during gait in individuals with diabetic neurop-athy. Clin. Biomech. 24, 687–692.

Saura, V., dos Santos, A.L.G., Ortiz, R.T., Parisi, M.C., Fernandes, T.D., Nery, M., 2010. Predic-tive factors of gait in neuropathic and non-neurophatic diabetic patients. Acta Ortop.Bras. 18, 148–151.

Savelberg, H.H., Schaper, N.C., Willems, P.J., de lange, T.L., Meijer, K., 2009. Redistributionof joint moments is associated with changed plantar pressure in diabeticpolyneuropathy. BMC Musculoskelet. Disord. 10, 16.

Savelberg, H.H.C.M., Ilgin, D., Angin, S., Willems, P.J.B., Schaper, N.C., Meijer, K., 2010.Prolonged activity of knee extensors and dorsal flexors is associated with adap-tations in gait in diabetes and diabetic polyneuropathy. Clin. Biomech. 25,468–475.

Sawacha, Z., Cristoferi, G., Guarneri, G., Corazza, S., Dona, G., Denti, P., et al., 2009a.Characterizing multisegment foot kinematics during gait in diabetic foot patients.J. Neuroeng. Rehabil. 6, 37.

Sawacha, Z., Gabriella, G., Cristoferi, G., Guiotto, A., Avogaro, A., Cobelli, C., 2009b. Diabeticgait and posture abnormalities: a biomechanical investigation through three dimen-sional gait analysis. Clin. Biomech. 24, 722–728.

Sawacha, Z., Guarneri, G., Cristoferi, G., Guiotto, A., Avogaro, A., Cobelli, C., 2012a. Integratedkinematics–kinetics–plantar pressure data analysis: a useful tool for characterizingdiabetic foot biomechanics. Gait Posture 1, 20–26.

Sawacha, Z., Spolaor, F., Guarneri, G., Contessa, P., Carraro, E., Venturin, A., et al., 2012b.Abnormal muscle activation during gait in diabetes patients with and without neu-ropathy. Gait Posture 35, 101–105.

Shenoy, A.M., 2012. Guidelines in practice: treatment of painful diabetic neuropathy. Con-tinuum Lifelong Learn. Neurol. 18, 192–198.

Singh, N., Armstrong, D., Lipsky, B., 2005. Preventing foot ulcers in patients with diabetes.JAMA 293, 217–228.

Stroup, D.F., Berlin, J.A., Morton, S.C., et al., 2000. Meta-analysis of observational studies inepidemiology: a proposal for reporting. JAMA 283, 2008–2012.

845M. Fernando et al. / Clinical Biomechanics 28 (2013) 831–845

Sutherland, D.E., 2001. The evolution of clinical gait analysis part l: kinesiological EMG.Gait Posture 1, 61–70.

Sutherland, D.H., 2002. The evolution of clinical gait analysis. Part II kinematics. Gait Pos-ture 16, 159–179.

Sutherland, D.H., 2005. The evolution of clinical gait analysis part III—kinetics and energyassessment. Gait Posture 21, 447–461.

Tesfaye, S., Selvarajah, D., 2012. Advances in the epidemiology, pathogenesis andmanagement of diabetic peripheral neuropathy. J. Diabetes Metab. Res. Rev. 1,8–14.

Turner, D.E., Helliwell, P.S., Burton, A.K., Woodburn, J., 2007. The relationship betweenpassive range of motion and range of motion during gait and plantar pressure mea-surements. Diabet. Med. 24, 1240–1246.

Uccioli, L., Caselli, A., Giacomozzi, C., Macellari, V., Giurato, L., Lardieri, L., et al., 2001. Pat-tern of abnormal tangential forces in the diabetic neuropathic foot. Clin. Biomech. 16,446–454.

Valensi, P., Giroux, C., Seeboth-Ghalayini, B., Attali, J.-R., 1997. Diabetic peripheralneuropathy: effects of age, duration of diabetes, glycemic control, and vascular fac-tors. J. Diabetes Complications 11, 27–34.

Van deursen, R., 2004. Mechanical loading and off-loading of the plantar surface of thediabetic foot. Clin. Infect. Dis. 39, 87–91.

Van dieren, S., Beulens, J.W., van der schouw, Y.T., Grobbee, D.E., Neal, B., 2010. The globalburden of diabetes and its complications: an emerging pandemic. Eur. J. Cardiovasc.Prev. Rehabil. 17 (Suppl. 1), s3–s8.

Winter, D.A., 1980. Overall principle of lower limb support during stance phase of gait.J. Biomech. 13, 923–927.

Wrobel, J.S., Najafi, B., 2010. Diabetic foot biomechanics and gait dysfunction. J. DiabetesSci. Technol. 4, 833–845.

Yavuzer, G., Yetkin, I., Toruner, F., Koca, N., Bolukbas, N., 2006. Gait deviations of patientswith diabetes mellitus: looking beyond peripheral neuropathy. Eur. Medicophys. 42,127–133.